The Nomadik clocksource driver has had a bad define making it
impossible to use it for sched_clock() for a while. Fix this
and also enable it for the Nomadik.
Signed-off-by: Linus Walleij <[email protected]>
---
arch/arm/mach-nomadik/Kconfig | 1 +
drivers/clocksource/nomadik-mtu.c | 4 ++--
2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/arch/arm/mach-nomadik/Kconfig b/arch/arm/mach-nomadik/Kconfig
index 954f0e3..8c1ccd4 100644
--- a/arch/arm/mach-nomadik/Kconfig
+++ b/arch/arm/mach-nomadik/Kconfig
@@ -5,6 +5,7 @@ config ARCH_NOMADIK
select ARM_AMBA
select ARM_VIC
select CLKSRC_NOMADIK_MTU
+ select CLKSRC_NOMADIK_MTU_SCHED_CLOCK
select COMMON_CLK
select CPU_ARM926T
select GENERIC_CLOCKEVENTS
diff --git a/drivers/clocksource/nomadik-mtu.c b/drivers/clocksource/nomadik-mtu.c
index 071f6ea..e405531 100644
--- a/drivers/clocksource/nomadik-mtu.c
+++ b/drivers/clocksource/nomadik-mtu.c
@@ -67,7 +67,7 @@ static u32 clk_prescale;
static u32 nmdk_cycle; /* write-once */
static struct delay_timer mtu_delay_timer;
-#ifdef CONFIG_NOMADIK_MTU_SCHED_CLOCK
+#ifdef CONFIG_CLKSRC_NOMADIK_MTU_SCHED_CLOCK
/*
* Override the global weak sched_clock symbol with this
* local implementation which uses the clocksource to get some
@@ -233,7 +233,7 @@ void __init nmdk_timer_init(void __iomem *base, int irq)
pr_err("timer: failed to initialize clock source %s\n",
"mtu_0");
-#ifdef CONFIG_NOMADIK_MTU_SCHED_CLOCK
+#ifdef CONFIG_CLKSRC_NOMADIK_MTU_SCHED_CLOCK
setup_sched_clock(nomadik_read_sched_clock, 32, rate);
#endif
--
1.8.1.4
ARM SoC folks:
On Sat, Apr 20, 2013 at 3:08 PM, Linus Walleij <[email protected]> wrote:
> The Nomadik clocksource driver has had a bad define making it
> impossible to use it for sched_clock() for a while. Fix this
> and also enable it for the Nomadik.
>
> Signed-off-by: Linus Walleij <[email protected]>
This one should go to fixes ASAP, could you pick it to your fixes
branch? It could be tagged stable if desired - this only affects the
Nomadik which works just OK on jiffies so it's really no big regression.
The Ux500 does not use the MTU for sched_clock().
Yours,
Linus Walleij
On Sat, Apr 20, 2013 at 03:19:43PM +0200, Linus Walleij wrote:
> ARM SoC folks:
>
> On Sat, Apr 20, 2013 at 3:08 PM, Linus Walleij <[email protected]> wrote:
>
> > The Nomadik clocksource driver has had a bad define making it
> > impossible to use it for sched_clock() for a while. Fix this
> > and also enable it for the Nomadik.
> >
> > Signed-off-by: Linus Walleij <[email protected]>
>
> This one should go to fixes ASAP, could you pick it to your fixes
> branch? It could be tagged stable if desired - this only affects the
> Nomadik which works just OK on jiffies so it's really no big regression.
> The Ux500 does not use the MTU for sched_clock().
Ok, applied on top of next/multiplatform, seems to be what it's based on.
Tagging as stable is up to you as platform owner. :) Since it won't directly
apply to older kernels you might be better of submitting it to Greg separately.
-Olof
On Sat, Apr 27, 2013 at 2:12 AM, Olof Johansson <[email protected]> wrote:
> On Sat, Apr 20, 2013 at 03:19:43PM +0200, Linus Walleij wrote:
>> On Sat, Apr 20, 2013 at 3:08 PM, Linus Walleij <[email protected]> wrote:
>>
>> > The Nomadik clocksource driver has had a bad define making it
>> > impossible to use it for sched_clock() for a while. Fix this
>> > and also enable it for the Nomadik.
>> >
>> > Signed-off-by: Linus Walleij <[email protected]>
>>
>> This one should go to fixes ASAP, could you pick it to your fixes
>> branch? It could be tagged stable if desired - this only affects the
>> Nomadik which works just OK on jiffies so it's really no big regression.
>> The Ux500 does not use the MTU for sched_clock().
>
> Ok, applied on top of next/multiplatform, seems to be what it's based on.
>
> Tagging as stable is up to you as platform owner. :) Since it won't directly
> apply to older kernels you might be better of submitting it to Greg separately.
Oh it's not *that* critical. It is not regressing anything in a bad way,
jiffies work fine. But this is the proper way for the Nomadik long-term.
Yours,
Linus Walleij